|Home | About | Journals | Submit | Contact Us | Français|
Increased body size and physical inactivity are positively related to risk of several cancers, but only few epidemiologic studies have investigated body-mass index (BMI) and physical activity in relation to thyroid cancer. We examined the relations of BMI and physical activity to thyroid cancer in a prospective cohort of 484,326 United States men and women, followed from 1995/1996 to 2003. During 3,490,300 person-years of follow-up, we documented 352 newly incident cases of thyroid cancer. The multivariate relative risks (RR) of thyroid cancer for BMI values of 18.5 to 24.9 (reference), 25.0 to 29.9, and ≥30 kg/m2 were 1.0, 1.27, and 1.39 [95% confidence interval (CI)=1.05–1.85]. Adiposity predicted papillary thyroid cancers (RR comparing extreme BMI categories=1.47; 95%-CI=1.03–2.10) and, based on small numbers, suggestively predicted follicular thyroid cancers (RR=1.49; 95%-CI=0.79–2.82) and anaplastic thyroid cancers (RR=5.80; 95%-CI=0.99–34.19). No relation with BMI was noted for medullary thyroid cancers (RR=0.97; 95%-CI=0.27–3.43). The positive relation of BMI to total thyroid cancer was evident for men but not for women. However, the test of interaction (p=0.463) indicated no statistically significant gender difference. Physical activity was unassociated with thyroid cancer. The RRs of total thyroid cancer for low (reference), intermediate, and high level of physical activity were 1.0, 1.01, and 1.01 (95%-CI=0.76–1.34, p for trend=0.931), respectively. Our results support an adverse effect of adiposity on risk for developing total and papillary, and possibly follicular thyroid cancers. Based on only 15 cases, adiposity was unrelated to medullary thyroid cancers. Physical activity was unrelated to total thyroid cancer.
Thyroid cancer is the most common malignancy of the endocrine system. An estimated 37,340 new thyroid cancer cases (28,410 in women and 8,930 in men) were predicted to occur and 1,590 people (910 women and 680 men) to die from this disease in the United States (U.S.) in 2008.1 In recent years, incidence rates of thyroid cancer increased sharply. Much of this increase has occurred in small papillary cancers and has been more pronounced among women. While some of the rise is due to enhanced detection of sub-clinical thyroid tumors from greater use of ultrasound and fine needle biopsies,2 a recent analysis of U.S. Surveillance Epidemiology and End Results (SEER) data shows a smaller but statistically significant rise in tumors >2 cm and in cancers diagnosed at regional stage.3 Thus, improvements in diagnostic techniques do not appear to explain all of the increase in thyroid cancer incidence over time and additional factors such as adiposity may contribute.
The strongest known determinant of thyroid cancer is exposure to ionizing radiation. Other risk factors include a history of goiter or thyroid nodules. Positive associations between height, weight, and body mass index (BMI) and thyroid cancer, as well as certain hormonal and reproductive factors among women, such as late age at menarche, late age at first birth, a history of miscarriage of the first pregnancy, and oral contraceptive use have been reported, but these associations generally have been weak and have not always been consistent across studies. In contrast, dietary intakes of fish and cruciferous vegetables, and cigarette smoking have been inversely related to thyroid cancer.4, 5
Approximately two-thirds of U.S. adults are currently overweight (BMI 25.0–29.9 kg/m2) or obese (BMI≥30.0 kg/m2) 6 and over half do not meet the physical activity recommendations.7 Because epidemiologic data on BMI and physical activity in relation to thyroid cancer are limited, we undertook a prospective investigation of these issues in a large cohort including 352 incident cases of thyroid cancer.
The NIH-AARP Diet and Health Study was initiated in 1995–1996 when 566,402 members of AARP (formerly known as the American Association of Retired Persons) 50 to 71 years old and residing in one of six U.S. states (CA, FL, LA, NJ, NC, and PA) or two metropolitan areas (Atlanta, GA, and Detroit, MI) satisfactorily completed and returned a mailed questionnaire on their medical history, diet, weight, height, and physical activity habits.8 Because the AARP membership contains a greater proportion of men than women, approximately two-thirds of our cohort is men and one-third is women.
At baseline, we excluded subjects with diagnosed cancer other than non-melanoma skin cancer (n=52,561), those with missing data on body weight or height (n=12,620), those who were underweight (defined as BMI<18.5 kg/m2, n=5,217) and those with missing information on physical activity (n=5,283) or smoking (n=6,395). The remaining analytical cohort consisted of 484,326 individuals. The study was approved by the Special Studies Institutional Review Board (IRB) of the U.S. National Cancer Institute.
We identified incident cases of thyroid cancer by probabilistic linkage to the state cancer registries serving our cohort. In addition to the eight original states of our cohort, our cancer registry ascertainment area was recently expanded to include three additional states (TX, AZ, and NV) in order to ascertain cancer cases among participants who relocated to those states during follow-up. The North American Association of Central Cancer Registries (NAACCR) certifies all eleven cancer registries.9 Cases of fatal thyroid cancer and vital status of cohort members were identified through linkage to the Social Security Administration Death Master file and to the National Death Index (NDI) Plus, the latter of which provided verification of death and cause of death information. For matching purposes, we have virtually complete data on first and last name, address history, gender, and date of birth. Social Security number is available for 85% of our cohort members.
In a previous validation study, we compared data from cancer registries to self-reports and medical records and reported that our method of linkage to cancer registries identified approximately 90% of all cancer cases occurring in our cohort.10 Cancer sites were identified by anatomic site and histologic code of the International Classification of Disease for Oncology (ICD-O, second and third editions).11 The primary endpoint for the present analysis was total thyroid cancer (ICD-10 code 73.9). To address the potential impact of etiologic heterogeneity among histologic types of thyroid cancer on the observed relations in our study, we also evaluated thyroid cancer according to histologic types, which included cancers with the following ICD-O-3 morphology: papillary (8050, 8052, 8130, 8260, 8340, 8341, 8342, 8343, 8344, 8450, 8452), follicular (8290, 8330, 8331, 8332, 8335), medullary (8345, 8346, 8510), and anaplastic (8021). There were 23 cases of thyroid cancer with missing information on histology. Of 329 thyroid cancer cases with complete histology, 244 cases (74%) were papillary, 62 cases (19%) were follicular, 15 cases (5%) were medullary, and 8 cases (2%) were anaplastic thyroid cancers.
Information on weight and height was requested on the baseline questionnaire and was used to calculate BMI and to divide subjects into three BMI categories defined by the World Health Organization as normal weight (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), and obesity (≥30.0 kg/m2).12
We inquired about the average frequency (never; rarely; 1 to 3 times per month; 1 to 2 times per week; 3 to 4 times per week; and 5 or more times per week) during the past year that participants engaged in activities of any type that lasted 20 minutes or more and caused either increases in breathing or heart rate or working up a sweat. Our physical activity assessment corresponds to the American College of Sports Medicine (ACSM) guidelines that recommend at least 20 minutes of continuous vigorous exercise three times per week for improving cardio-respiratory fitness.13 We created three categories of physical activity to ensure adequate case numbers in each physical activity category across thyroid cancer histologic types: low (less than 3 times per week); intermediate (3 to 4 times per week); and high (5 or more times per week). A physical activity assessment similar to the one used in our cohort demonstrated good inter-rater reliability (percentage agreement=0.76; kappa=0.53) and reasonable validity (percentage agreement=0.71; kappa=0.40) as assessed by a computer science and applications (CSA) activity monitor.14 In addition, our physical activity assessment predicts lower risk of mortality from coronary heart disease.15
Follow-up time extended from the scan date of the baseline questionnaire to the diagnosis date of thyroid cancer, leaving the registry ascertainment area, death, or December 31, 2003, whichever occurred first. We used Cox proportional hazards regression16 with person-time as the underlying time metric to estimate relative risks (RR) of developing thyroid cancer by levels of BMI and physical activity. Alternative analyses using age or calendar time as the underlying time metric generated virtually identical results (data not shown). We assessed and verified no violation of the proportional hazards assumption. The association between BMI and thyroid cancer was evaluated in two models. One model was adjusted for age (continuous) and gender (female; male). The second model was additionally adjusted for physical activity (low; intermediate; high), race/ethnicity (Caucasian; non-Caucasian), education (less than college; college graduate or postgraduate), smoking status (never; former; current), current alcohol use (yes; no); and oral contraceptive use among women (ever; never). We also considered the following variables, but did not retain those variables in our final models because their inclusion had no impact on the relations of BMI or physical activity on thyroid cancer: family history of cancer, age at menarche, age at first birth, menopausal hormone use, and dietary intakes of fruit and vegetables, cruciferous vegetables specifically, and fish. The relation of physical activity to thyroid cancer was estimated in two models that were similar to the BMI models, with the exception that physical activity was replaced by BMI in the second model.
Tests of linear trend across categories of the main exposures were conducted by entering the median values of exposures as single continuous variables in the multivariate model, the coefficients for which were assessed using a Wald test. For BMI we used median values of 21.7, 27.5, and 32.5 kg/m2 and for physical activity we used median values of 1.5, 3.5, and 5.5 times per week.
To investigate whether the relations of BMI or physical activity to thyroid cancer risk were modified by other potential risk factors for thyroid cancer, we performed tests for multiplicative interaction using likelihood-ratio tests. We also examined the relations of BMI and physical activity to thyroid cancer within strata of potential effect modifiers. All RRs are displayed with 95% confidence intervals (CI), and reported p values are based on two-sided tests. All statistical analyses were carried out using SAS release 8.2 (SAS Institute, Cary, NC).
During 3,490,300 person years of follow-up, we identified 352 cases of thyroid cancer, of which 171 were diagnosed in men and 181 in women. Participants with high BMI were less physically active, they were more likely to have a history of diabetes, they had a lower education level, and they consumed less alcohol than those with normal BMI. Subjects with high physical activity were leaner, they were less likely to have a history of diabetes, they were more educated, and they consumed slightly more alcohol than less active subjects. At baseline, participants with high BMI or those with high physical activity levels were less likely to currently smoke, but they were more likely to have formerly smoked than their lean or less active counterparts (Table 1).
Participants who were overweight or obese had a greater risk of thyroid cancer than normal weight subjects. After controlling for age and gender, individuals with BMI levels of 18.5–24.9 (reference), 25.0–29.9, and ≥ 30 kg/m2 had RRs of 1.0, 1.27, and 1.41 (95%-CI=1.06–1.86; p for trend=0.004) (Table 2). Upon additional adjustment for multiple variables, the positive association was virtually unaltered (RR comparing extreme BMI categories=1.39; 95%-CI=1.05–1.85; p for trend=0.007). When we evaluated BMI in relation to thyroid cancer according to histologic types, statistical power was reduced, but positive associations remained apparent for papillary thyroid cancers. Based on much smaller numbers, similar associations were observed for follicular and anaplastic thyroid cancers, whereas no relation with BMI was evident for medullary thyroid cancers.
To examine whether the relation between BMI and total thyroid cancer varied across strata defined by gender, age, race/ethnicity, history of screening for any cancer, history of diabetes, education, physical activity, smoking status, and alcohol use, we fit stratified models according to levels of those variables (Table 3). We observed a positive association between BMI and thyroid cancer in virtually all subgroups. The relation of BMI to thyroid cancer was stronger among men than women and Caucasians compared with non-Caucasians, but the tests of interaction indicated no statistically significant difference between strata of gender or race/ethnicity.
In a secondary analysis, we investigated the association between BMI at age 18 years and total thyroid cancer. As compared with participants who were normal weight (BMI=18.5–24.9 kg/m2) at age 18 years, the multivariate RR of thyroid cancer for those who were overweight (BMI=25.0–29.9 kg/m2) at age 18 years was 1.50 (95% CI=0.97–2.33), suggesting that adiposity at a younger age also conferred increased risk of thyroid cancer. In our dataset, there were only two cases of thyroid cancer that were obese (BMI≥30.0 kg/m2) at age 18 years, thus we do not present risk estimates relating obesity at age 18 years to thyroid cancer.
Physical activity was unrelated to risk of total thyroid cancer. Subjects who reported engaging in increasing levels of physical activity [low (reference), intermediate, high] had multivariate RRs of 1.0, 1.01, and 1.01 (95%-CI=0.76–1.34; p for trend<0.931) (Table 4). Likewise, no association with physical activity emerged when we examined thyroid cancer according to histologic types. We did note a positive relation between physical activity and anaplastic thyroid cancers but that analysis was based on only 4 exposed cases.
Null associations between physical activity and total thyroid cancer were found for all subgroup analyses defined by gender, age, race/ethnicity, history of screening for any cancer, history of diabetes, education, physical activity, smoking status, and alcohol use and formal tests of interaction were not statistically significant (Table 5).
In this study of nearly half a million individuals followed for up to eight years, we found a modest but graded positive association between BMI and thyroid cancer. Specifically, overweight was related to a suggestive increase of nearly 30% in the risk of thyroid cancer and obesity was associated with an increase of 40% in the risk of thyroid cancer when compared to normal weight. The apparent adverse effect of adiposity on risk for thyroid cancer was evident for papillary and possibly follicular cancers, whereas no relation with BMI was seen for medullary thyroid cancers. The positive relation of BMI to total thyroid cancer was evident for men but not for women and was evident for Caucasians but not non-Caucasians. However, interaction tests suggested that these differences were not statistically significant. The positive relation of BMI to thyroid cancer was independent of other known or suspected risk factors for thyroid cancer we were able to control for, including age, smoking, and diet suggesting that avoidance of adiposity may play an important independent role in the prevention of thyroid cancer.
Our data showing a modest positive relation between BMI and total thyroid cancer is compatible with results from a recent meta-analysis17 of prospective studies on this topic. That meta-analysis17 included 3,587 thyroid cancer cases and found that each 5 kg/m2 increase in BMI was associated with a pooled RR of 1.14 (95% CI=1.06–1.23) for women and a pooled RR of 1.34 (95% CI=1.04–1.70) for men. Our findings with respect to histologic types of thyroid cancer are also consistent with a recent large prospective investigation from Norway18 that observed positive associations with BMI for papillary, follicular, and anaplastic thyroid cancers, whereas null or inverse associations with BMI were noted for medullary thyroid cancers. Heterogeneity in the relations of BMI to non-medullary versus medullary thyroid cancers may be explained by the distinct etiologies of thyroid cancer histologic types.19
While several prospective investigations support a positive relation of BMI to thyroid cancer,18, 20, 21 two cohort studies,22, 23 one22 of which included both women and men whereas the other23 involved only men, reported a null association between BMI and thyroid cancer. Retrospective data24 from an Italian cohort of 1,333 morbidly obese patients aged 21 to 70 years who were referred for bariatric surgery revealed that thyroid cancer was the second most frequent site of cancer prevalence (18.6%) among those who presented with a cancer diagnosis at the time of referral, adding further evidence for a link between adiposity and thyroid cancer risk, although the latter finding could in part be due to selection bias because obese patients are seen by doctors more frequently than non-obese patients.
Although results differ somewhat among studies, case-control studies generally support a positive association between BMI and thyroid cancer. A pooled analysis of case-control investigations of BMI and thyroid cancer25 including 2,473 cases reported a pooled OR of 1.2 (95% CI=1.0–1.4) comparing extreme tertiles of BMI among women, but found no relation between BMI and thyroid cancer among men (pooled OR=1.0; 95% CI=0.8–1.4). The authors of the pooled analysis25 speculated that one possible reason for the weaker results among men than women was the considerably smaller sample size among men. Although not statistically significantly different, results from our study with an equal gender distribution of cases appeared stronger for men than women.
Several case-control studies have been published since the above-mentioned pooled analysis. A recent case-control study26 from New Caledonia observed an odds ratio (OR) of thyroid cancer of 1.85 (95% CI=1.02–3.35) for BMI of ≥35 kg/m2 compared to normal BMI among women, however, no relation in men (OR=1.04; 95% CI=0.28–3.79) comparing extreme categories of BMI. One case-control study27 among U.S. women found no relation between BMI and thyroid cancer but data were not presented. However, that study27 reported ORs of total and papillary thyroid cancer of 1.6 (95% CI=0.9–3.0) and 1.4 (95% CI=0.7–2.6), respectively, for weight gain of greater than 21 pounds between age 18 years and the reference age. One case-control study28 focusing on diet and thyroid cancer showed descriptive data suggesting a slightly higher BMI among thyroid cancer cases (25.4 kg/m2) than controls (24.3 kg/m2) but that comparison was based on an unadjusted analysis and the difference was only borderline statistically significant (p=0.05).
In contrast to the generally positive results for body mass in our and other studies, we found that physical activity was not associated with risk of total thyroid cancer in our cohort, although there was some suggestion of a positive relationship with the anaplastic cancer subtype. Available epidemiologic data regarding the relation of physical activity to thyroid cancer are sparse. One early case-control study29 found no statistically significant relation of occupational physical activity to thyroid cancer risk. A more recent case-control study30 among women reported an OR of papillary thyroid cancer of 0.76 (95% CI=0.59–0.98) for regular exercise during the two years before diagnosis compared to no regular exercise during that time period. Thus, the relation of physical activity to thyroid cancer remains unresolved.
Advantageous features of our study include its prospective design, a high follow-up rate, large sample size with a relatively sizeable number of cases, available data on thyroid cancer histologic types, and information on several potential thyroid cancer risk factors. Measurement error in assessing body size is not likely to be a concern because self-reported weight and height have been found to be highly accurate.31
Potential limitations of our study involve imprecise estimation of physical activity levels owing to the self-reported nature of our activity assessment,32 but, validation studies comparing physical activity assessments similar to those used in our cohort with referent methods indicate that the reliability and validity of our instrument is comparable to self-reported measures used in similar studies.33
A further potential shortcoming of our study is that we lacked sufficient numbers of thyroid cancer cases among non-Caucasians to evaluate whether the relation between BMI and total thyroid cancer varies according to ethnic or racial group. Some evidence suggests that associations with BMI differ by ethnic group. For example, a large prospective study21 of white and black U.S. veterans compared obese to non-obese men and found an RR of 1.92 (95% CI=1.09–3.40) for black men, whereas the RR for white men was less pronounced (corresponding RR=1.40; 95% CI=1.09–1.81). A particularly strong positive relation of BMI to thyroid cancer was observed in a case-control study among Melanesian women,26 showing an OR of 2.40 (95% CI=1.19–4.84) for individuals with a BMI of 30.-34.9 kg/m2 compared to those with a BMI of 18.5–24.9 kg/m2. A prospective study of Korean men20 found a RR of thyroid cancer of 2.23 (95% CI=1.40–3.55) comparing men with a BMI of 27.0–29.9 kg/m2 to those with a BMI of 18.5–22.9 kg/m2.
Our study lacked information on certain determinants of thyroid cancer, such as history of radiation exposure. Such factors could potentially confound the relation of adiposity or physical activity to thyroid cancer. Nonetheless, we carefully controlled for numerous potential confounding variables, including age, gender, race/ethnicity, education, smoking status, alcohol use, and oral contraceptive use among women. Moreover, inclusion of these risk factors in the models had little impact on the risk estimates, suggesting that regulation of these variables explains only a small portion of the apparent adverse effects of adiposity on thyroid cancer risk.
It has been hypothesized that the observed positive relation of adiposity to thyroid cancer risk may be due to detection bias because of more frequent examinations of the thyroid gland among overweight/obese than lean individuals.25 We lacked information on thyroid carcinoma size among cases in our study, which would have enabled us to uncover potential detection bias by focusing on relations with small sized thyroid carcinomas, which are more prone to be detected with more frequent examinations of the thyroid gland.34 Other methodologic biases do not likely explain the positive association with BMI seen in our data. Specifically, information on weight, height, and other exposures was collected prior to thyroid cancer diagnoses, which precluded bias ascribable to differential recall of weight, height, or physical activity by study participants with and without a subsequent diagnosis of thyroid cancer.
Biological mechanisms through which BMI may relate to thyroid cancer are speculative. Excess adiposity is related to increased insulin production, and insulin may enhance tumor growth by increasing free insulin-like growth factor (IGF)-1, which in turn stimulates cell proliferation and suppresses apoptosis and has been positively linked to thyroid cancer.35 Although hyperinsulinemia per se has not been implicated in thyroid carcinogenesis, hyperglycemia is directly associated with thyroid cancer.36 Among postmenopausal women, increased risk of thyroid cancer associated with adiposity may also be explained by increased production of endogenous estrogens in the adipose tissue37 because estrogens potentially promote thyroid carcinogenesis38 and have been found to be positively related to thyroid cancer.39 In addition, adiposity may increase thyroid cancer risk through its effects on thyroid stimulating hormone (TSH),40 which represents an independent predictor of thyroid malignancy.41
In summary, this study provides evidence for an adverse effect of adiposity on thyroid cancer risk. The positive relation with BMI was evident for papillary, and possibly follicular histologic subtypes of thyroid cancers. In contrast, based on only 15 cases, BMI did not appear to be associated with medullary thyroid cancers. The positive association between BMI and total thyroid cancer was evident for men but not for women, although the test of interaction indicated no statistically significant gender difference. No relation was observed between physical activity and total thyroid cancer although there was some suggestion of a positive relationship with the anaplastic cancer subtype. Future mechanistic studies should clarify potential biological mechanisms underlying the positive relation of adiposity to thyroid cancer risk.